Question

我将一个大对象数组序列化为json http响应流。现在我想一次一个地从流中反序列化这些对象。是否有任何c＃库可以让我这样做？我看过json.net，但似乎我必须立即反序列化完整的对象数组。

[{large json object},{large json object}.....]

澄清：我想一次从流中读取一个json对象并反序列化。

Answer 1

为了逐步读取JSON，您需要将JsonTextReader与StreamReader结合使用。但是，您不一定要从阅读器手动读取所有JSON。您应该能够利用Linq-To-JSON API从阅读器加载每个大对象，以便您可以更轻松地使用它。

举个简单的例子，假设我有一个看起来像这样的JSON文件：

[
  {
    "name": "foo",
    "id": 1
  },
  {
    "name": "bar",
    "id": 2
  },
  {
    "name": "baz",
    "id": 3
  }
]

从文件中逐步读取它的代码可能类似于以下内容。（在您的情况下，您将使用您的响应流替换FileStream。）

using (FileStream fs = new FileStream(@"C:\temp\data.json", FileMode.Open, FileAccess.Read))
using (StreamReader sr = new StreamReader(fs))
using (JsonTextReader reader = new JsonTextReader(sr))
{
    while (reader.Read())
    {
        if (reader.TokenType == JsonToken.StartObject)
        {
            // Load each object from the stream and do something with it
            JObject obj = JObject.Load(reader);
            Console.WriteLine(obj["id"] + " - " + obj["name"]);
        }
    }
}

上述输出如下：

1 - foo
2 - bar
3 - baz

Answer 2

我简化了解析器/解串器的一个示例/测试，以更直接地回答这个问题的用例。

以下是测试数据：

https://github.com/ysharplanguage/FastJsonParser/tree/master/JsonTest/TestData

（参见fathers.json.txt）

以下是示例代码：

    using System;
    using System.Collections.Generic;
    using System.IO;
    using System.Linq;
    using System.Text;

    // Our stuff
    using System.Text.Json;

//...

    public class FathersData
    {
        public Father[] fathers { get; set; }
    }

    public class Someone
    {
        public string name { get; set; }
    }

    public class Father : Someone
    {
        public int id { get; set; }
        public bool married { get; set; }
        // Lists...
        public List<Son> sons { get; set; }
        // ... or arrays for collections, that's fine:
        public Daughter[] daughters { get; set; }
    }

    public class Child : Someone
    {
        public int age { get; set; }
    }

    public class Son : Child
    {
    }

    public class Daughter : Child
    {
        public string maidenName { get; set; }
    }

//...

    static void FilteredFatherStreamTestSimplified()
    {
        // Get our parser:
        var parser = new JsonParser();

        // (Note this will be invoked thanks to the "filters" dictionary below)
        Func<object, object> filteredFatherStreamCallback = obj =>
        {
            Father father = (obj as Father);
            // Output only the individual fathers that the filters decided to keep (i.e., when obj.Type equals typeof(Father)),
            // but don't output (even once) the resulting array (i.e., when obj.Type equals typeof(Father[])):
            if (father != null)
            {
                Console.WriteLine("\t\tId : {0}\t\tName : {1}", father.id, father.name);
            }
            // Do not project the filtered data in any specific way otherwise,
            // just return it deserialized as-is:
            return obj;
        };

        // Prepare our filter, and thus:
        // 1) we want only the last five (5) fathers (array index in the resulting "Father[]" >= 29,995),
        // (assuming we somehow have prior knowledge that the total count is 30,000)
        // and for each of them,
        // 2) we're interested in deserializing them with only their "id" and "name" properties
        var filters = 
            new Dictionary<Type, Func<Type, object, object, int, Func<object, object>>>
            {
                // We don't care about anything but these 2 properties:
                {
                    typeof(Father), // Note the type
                    (type, obj, key, index) =>
                        ((key as string) == "id" || (key as string) == "name") ?
                        filteredFatherStreamCallback :
                        JsonParser.Skip
                },
                // We want to pick only the last 5 fathers from the source:
                {
                    typeof(Father[]), // Note the type
                    (type, obj, key, index) =>
                        (index >= 29995) ?
                        filteredFatherStreamCallback :
                        JsonParser.Skip
                }
            };

        // Read, parse, and deserialize fathers.json.txt in a streamed fashion,
        // and using the above filters, along with the callback we've set up:
        using (var reader = new System.IO.StreamReader(FATHERS_TEST_FILE_PATH))
        {
            FathersData data = parser.Parse<FathersData>(reader, filters);

            System.Diagnostics.Debug.Assert
            (
                (data != null) &&
                (data.fathers != null) &&
                (data.fathers.Length == 5)
            );
            foreach (var i in Enumerable.Range(29995, 5))
                System.Diagnostics.Debug.Assert
                (
                    (data.fathers[i - 29995].id == i) &&
                    !String.IsNullOrEmpty(data.fathers[i - 29995].name)
                );
        }
        Console.ReadKey();
    }

其余位可在此处获取：

https://github.com/ysharplanguage/FastJsonParser

“HTH，

Answer 3

这是我的解决方案（从不同来源合并，但主要基于"Checking for Protocol Conformance"解决方案）将巨大的JSON文件（对象数组）转换为任何通用对象的XML文件。

JSON看起来像这样：

   {
      "Order": [
          { order object 1},
          { order object 2},
          {...}
          { order object 10000},
      ]
   }

输出XML：

<Order>...</Order>
<Order>...</Order>
<Order>...</Order>

C＃代码：

XmlWriterSettings xws = new XmlWriterSettings { OmitXmlDeclaration = true };
using (StreamWriter sw = new StreamWriter(xmlFile))
using (FileStream fs = new FileStream(jsonFile, FileMode.Open, FileAccess.Read))
using (StreamReader sr = new StreamReader(fs))
using (JsonTextReader reader = new JsonTextReader(sr))
{
    //sw.Write("<root>");
    while (reader.Read())
    {
        if (reader.TokenType == JsonToken.StartArray)
        {
            while (reader.Read())
            {
                if (reader.TokenType == JsonToken.StartObject)
                {
                    JObject obj = JObject.Load(reader);
                    XmlDocument doc = JsonConvert.DeserializeXmlNode(obj.ToString(), "Order");
                    sw.Write(doc.InnerXml); // a line of XML code <Order>...</Order>
                    sw.Write("\n");
                    //this approach produces not strictly valid XML document
                    //add root element at the beginning and at the end to make it valid XML                                
                }
            }
        }
    }
    //sw.Write("</root>");
}

Answer 4

使用Cinchoo ETL-一个开放源代码库，您可以有效地解析低内存占用的大型JSON。由于对象是在基于流的拉模型中构造并返回的，所以

using (var p = new ChoJSONReader(** YOUR JSON FILE **))
{
            foreach (var rec in p)
            {
                Console.WriteLine($"Name: {rec.name}, Id: {rec.id}");
            }
}

有关更多信息，请访问codeproject文章。

希望有帮助。

一次反序列化一个项目的json数组流

4 个答案: